Method of Performance Analysis of Time-Critical Applications Using DB-Nets

Anton Mikhailovich Rigin,Sergey Andreevich Shershakov

doi:10.15514/ispras-2021-33(3)-9

Anton Mikhailovich Rigin, Sergey Andreevich Shershakov

Open Access

https://doi.org/10.15514/ispras-2021-33(3)-9

Copy DOI

Abstract

These days, most of time-critical business processes are performed using computer technologies. As an example, one can consider financial processes including trading on stock exchanges powered by electronic communication protocols such as the Financial Information eXchange (FIX) Protocol. One of the main challenges emerging with such processes concerns maintaining the best possible performance since any unspecified delay may cause a large financial loss or other damage. Therefore, performance analysis of time-critical systems and applications is required. In the current work, we develop a novel method for a performance analysis of time-critical applications based on the db-net formalism, which combines the ability of colored Petri nets to model a system control flow with the ability to model relational database states. This method allows to conduct a performance analysis for time-critical applications that work as transactional systems and have log messages which can be represented in the form of table records in a relational database. One of such applications is a FIX protocol-based trading communication system. This system is used in the work to demonstrate applicability of the proposed method for time-critical systems performance analysis. However, there are plenty of similar systems existing for different domains, and the method can also be applied for a performance analysis of these systems. The software prototype is developed for testing and demonstrating abilities of the method. This software prototype is based on an extension of Renew software tool, which is a reference net simulator. The testing input for the software prototype includes a test log with FIX messages, provided by a software developer of testing solutions for one of the global stock exchanges. An application of the method for quantitative analysis of maximum acceptable delay violations is presented. The developed method allows to conduct a performance analysis as a part of conformance checking of a considered system. The method can be used in further research in this domain as well as in testing the performance of real time-critical software systems.

Highlights

Nowadays, most of time-critical business processes are performed using computer technologies
We develop a novel method for a performance analysis of time-critical applications based on the db-net formalism, which combines the ability of colored Petri nets to model a system control flow with the ability to model relational database states
This method allows to conduct a performance analysis for time-critical applications that work as transactional systems and have log messages which can be represented in the form of table records in a relational database

Summary

Introduction

Most of time-critical business processes are performed using computer technologies. Different financial processes including trading on stock exchanges can demand strict performance requirements. In the previous century, trading on stock exchanges was primarily performed through phone calls and with use of paper-based order books [1]. Working this way did not allow traders to compete for the best price that is generally offered during very short period. This was the reason of beginning of automatization of trading on stock exchanges.

Objectives

Methods

Conclusion